Popular tools

Theme

Jitter Test

See how evenly your packets arrive: real UDP at the rates calls and games use, timed idle and while your line is busy.

RFC 3550 jitter, as call apps compute it · UDP to Cloudflare's nearest data centre · nothing is stored
Profile

3,000 UDP packets through Cloudflare's nearest data centre and back · 1 min · nothing is stored

Jitter over timeExample: a made-up minute

worst 15 s: 35.9 ms15304560Teams30 ms010 s20 s30 s40 s50 s60 s

Every packet, as it landedExample: a made-up minute

  • On time 2,632
  • Late 103
  • Lost 3
  • Out of order 262
  • Duplicate 0
PacketsBurstsUploadUpDownloadDownDelay ms3 lostnone–
Each column is one second: its packets top to bottom. Point at a second, or focus the strip and use the arrow keys.

How does this jitter test work?

This network jitter test sends small UDP packets through Cloudflare's nearest data centre and back, 50 a second by default, the rate a voice call sends them, and times every one on one clock.

Your jitter is RFC 3550's interarrival jitter: how much the delay changes from one packet to the next, as a running average. It is the number WebRTC reports and call apps use, so it compares with what Teams or Zoom see.

An average can hide a bad moment, so the test also finds the worst 15 seconds of the run, the window Microsoft Teams' network target uses. With the line busy, it measures jitter while downloading and while uploading too.

A 1-minute jitter test, second by seconda made-up line
IdleDownloadingUploading worst 15 s: 35.9 ms 02550 ms Teams30 ms

Whole run 13.5 ms; worst 15 s 35.9 ms from 45 s, while uploading: above Teams' 30 ms.

Jitter test result: RFC 3550 jitter, the worst 15 seconds, Teams' and Zoom's limits, and jitter idle and under loadJitter test result: RFC 3550 jitter, the worst 15 seconds, Teams' and Zoom's limits, and jitter idle and under loadYour jitterRFC 3550 jitter over the whole run: the number call apps compute from the same packets.The worst 15 secondsAn average can hide a bad moment: this is the worst 15-second window, the one Microsoft Teams' target uses.The limitsMicrosoft Teams' and Zoom's published jitter limits, each checked against the run.Under loadJitter idle, while downloading and while uploading: whether a busy line makes it worse.
A 1-minute test on a made-up line: steady while idle, a spike while uploading. Select a number to see what each part shows.

What is jitter?

Jitter is how unevenly packets arrive: the change in delay from one packet to the next. A call sends a packet about every 20 ms; jitter is how far from every 20 ms they turn up.

The app at the other end holds incoming packets in a jitter buffer so it can play them out smoothly. A packet that arrives later than the buffer allows is thrown away, and the app fills its gap with sound it makes up itself. Microsoft says Teams absorbs some jitter this way; once jitter is more than the buffer covers, the speaker's voice sounds robotic.

Jitter isn't the same as packets arriving out of order (that's reordering). But when the delay changes by more than the gap between two packets, the later one overtakes, so high jitter often brings reordering too. This test counts them separately.

RFC 3550 interarrival jitterJ = J + (|D| − J) / 16
SentArrivedChange DJitter J 0 ms1820 ms21+340 ms19−260 ms26+780 ms18−8100 ms30+12120 ms19−11140 ms20+1160 ms22+22.31

Transit times in ms. Each change moves J a sixteenth of the way, so one spike lifts it a little and steady packets bring it back down.

Cloudflare's speed test averages the change between consecutive pings, Waveform and Ookla the spread around the mean: the same line gives three different jitter figures.

What is a good jitter?

For calls, under 30 ms in any 15 seconds meets Microsoft Teams' network target, and 40 ms or less meets Zoom's recommendation. Microsoft's call-quality guide counts audio jitter above 30 ms as poor.

The limits measure different things. Teams' target covers any 15 seconds of the path from your device to Microsoft's network. Zoom gives one figure for meetings and phone calls. ITU-T Y.1541 allows calls 50 ms of one-way delay variation, measured per minute, which can't be compared with a round-trip number.

So is 40 ms jitter bad? It is at Zoom's limit and above Teams'. Check the worst 15 seconds as well as the average: a line that is calm when idle can still pass 30 ms when someone starts an upload, as in the example above.

Published jitter limits, and what each one measures
WhoLimitMeasured over
Microsoft Teams (network target)under 30 msany 15 seconds, your device to Microsoft's network
Microsoft Teams (call quality)30 ms or lessa meeting's audio, on average
Zoom40 ms or lessmeetings and Zoom Phone
ITU-T Y.1541 (class 0, VoIP)50 ms or lessone way, per minute: not comparable to a round trip

Jitter vs latency (and packet loss)

Latency is how long a packet takes; jitter is how much that time changes; packet loss is how many never arrive. Microsoft describes each one's sound on a call: a long round trip gives a walkie-talkie effect, jitter a robotic voice, loss missing words.

They meet in the jitter buffer: a packet that arrives too late to be played is discarded, and the listener hears it as if it had been lost. So a call with a low ping can still break up when its jitter is high. The ping test measures latency to many regions; the packet loss test counts lost and late packets.

Same packets, two linesmade-up lines, a packet every 20 ms
Line A SentArrived Line B SentArrived
  • Line Aaverage round trip 41 msjitter 1.1 ms
  • Line Baverage round trip 26 msjitter 26.6 ms

Why do jitter tests disagree?

Because tools use different formulas, on different samples. On the same packets, a formula that measures change between packets and one that measures spread around the average can give very different numbers.

RFC 3550, used by call apps and WebRTC, keeps a running average of the change in delay from one packet to the next, smoothed so one spike moves it by only a sixteenth.

Cloudflare's speed test takes the plain average of the change between consecutive samples. On the same packets it lands close to RFC 3550: both measure change from one packet to the next.

Waveform and Ookla take the average distance of each sample from the mean. That also counts slow swings, a delay that drifts up and down over seconds, which the other two barely see.

Tools also test to different places, at different rates and for different lengths, so even one formula gives different numbers on different tests. This test shows its RFC 3550 number first and the other two beside it, from the same packets.

One line, three jitter formulasa made-up line
03060 ms average 33 ms

A bar every 200 ms; the numbers use all 750 packets.

  • RFC 3550call apps and WebRTC1.7 ms
  • Mean change between packetsthe formula of Cloudflare's speed test1.7 ms
  • Mean distance from the averagethe formula of Waveform and Ookla9.6 ms

What causes jitter?

Zoom's own definition names three causes: network congestion, timing drift and route changes. At home, congestion is usually the one you can fix.

A busy line. While a download or an upload fills your line, packets queue behind it and the round trip grows: Netflix calls this loaded latency. The test measures jitter while downloading and uploading to show it; the bufferbloat test grades the delay it adds.

Wi-Fi. Microsoft notes that wireless devices compete for bandwidth, that 2.4 GHz is often disturbed by other household devices, and its own call targets allow twice the poor-quality rate on 2.4 GHz as on a cable. Run the test on both to see the difference on your line.

The path further out. Congestion between your provider and the service, or a route change during the call. If your cable test is steady at home but calls still break up, the traceroute shows the path.

Three made-up lines, 15 seconds eachRFC 3550, second by second
  • A cable, nothing else running1.5 ms
  • A noisy Wi-Fi link8.7 ms
  • During a big upload35.0 ms

The dashed line is Microsoft Teams' 30 ms.

How to fix network jitter

Find where it comes from first: test on a cable, then on Wi-Fi, then with the line busy. Each result points at a different fix.

If jitter rises only while downloading or uploading, it's the queue in your router: turn on smart queue management (SQM), as the bufferbloat test explains, or pause large uploads during calls.

If it's high on Wi-Fi but low on a cable, move closer to the router and use 5 GHz, which Microsoft calls better suited to calls, or a cable. If it's high even on a cable with nothing else running, share the test's link with your provider: it shows the worst 15 seconds and when they happened.

Find where your jitter comes from

  1. 1

    Test on a cable, with nothing else running

    This is your line's own jitter. If it's already high, the cause is your provider or the path beyond it.

  2. 2

    Test again on Wi-Fi

    If jitter rises only here, move closer to the router, use 5 GHz or plug in for calls.

  3. 3

    Test with the load on

    If jitter rises while downloading or uploading, it's the queue in your router: turn on SQM, or pause big uploads during calls.

  4. 4

    Share the result

    Send the share link to your provider: it shows the worst 15 seconds and when they happened.

Jitter test: questions and answers

What is a good jitter?

For calls, under 30 ms in every 15 seconds meets Microsoft Teams' network target, and 40 ms or less meets Zoom's recommendation; Microsoft's call-quality guide counts audio jitter above 30 ms as poor. This test checks both limits and shows the worst 15 seconds as well as the average.

Is 40 ms jitter bad?

It's at Zoom's limit (40 ms or less) and above Microsoft Teams' (under 30 ms in any 15 seconds), so calls may break up. Check whether it's the average or a spike: the test shows the worst 15 seconds and when it happened, and whether it only rises while your line is busy.

Is 7 ms jitter bad?

No: it's well within both Microsoft Teams' limit (under 30 ms in every 15 seconds) and Zoom's (40 ms or less). Still look at the worst 15 seconds and at jitter while downloading and uploading: a line that's steady when idle can still spike when it's busy, which the test shows phase by phase.

How do I fix my internet jitter?

Test on a cable, then on Wi-Fi, then with the line busy. If jitter rises only under load, turn on smart queue management (SQM) on your router. If it's worse on Wi-Fi, use a cable or 5 GHz for calls. If it's high even on a cable with nothing else running, send the test's link to your provider.

Why is my jitter different on other tests?

Tools use different formulas: call apps and this test use RFC 3550, Cloudflare averages the change between samples, and Waveform and Ookla average the distance from the mean. They also test to different places, at different rates and for different lengths. This test shows the other formulas beside its own, from the same packets.

What is the difference between jitter and ping?

Ping (latency) is how long a packet takes to get there and back; jitter is how much that time changes from packet to packet. A call can have a low ping and still sound robotic if its jitter is high: packets that arrive later than the app's jitter buffer allows are thrown away, as if they had been lost.

Does Wi-Fi cause jitter?

It can: Microsoft notes that wireless devices compete for bandwidth and that 2.4 GHz is often disturbed by other household devices, and it calls 5 GHz better suited to calls. Run the test once on Wi-Fi and once on a cable to see how much yours adds.

How much data does the jitter test use?

The test packets themselves are small. Measuring under load moves up to 250 MB down and 100 MB up while your line is busy; turn the load off to test idle only. Nothing is stored on our side; past runs stay in your own browser.

Missing something?Suggest a feature or a new tool

Related tools